Is high expression of the chemokine receptor cxcr‐4 of predictive value for early relapse in childhood acute lymphoblastic leukaemia?

We were very interested in the paper of Crazzolara et al (2001), which indicated that overexpression of the chemokine receptor CXCR-4 on malignant acute lymphoblastic leukaemia (ALL) cells is associated with extramedullary organ infiltration. Therefore, we were prompted to analyse the expression of CXCR-4 on blast cells at diagnosis of 10 children with an extramedullary relapse (testis or central nervous system, with or without medullary relapse) [median white blood cells (WBC) at diagnosis: 30 × 109/l; median disease-free survival (DFS): 34 months]. The results were compared with the expression of CXCR-4 on leukaemic cells of 12 children with an isolated medullary relapse (median WBC: 106 × 109/l; DFS: 13 months) and of 25 leukaemic children with a favourable outcome (median follow-up 60 months, range 24–165, median WBC 38·3 × 109/l). Cells were stored in liquid nitrogen. Phenotypes, according to the criteria of the European Group for the Immunological Characterization of Leukaemia (EGIL) (Bene et al, 1995), were distributed as follows: three pre-B1, 26 pre-B2, 10 pre-B3, one mature B and seven T-cell acute lymphoblastic leukaemia (T-ALL). CXCR-4 expression [Mean Fluorescence Intensity (MFI), percentage of positive cells] was detected by indirect immunofluorescence using an antibody against CXCR-4 (clone 12G5, R & D Systems, UK) as previously described (Sbaa-Ketata et al, 2001). Immunofluorescence analysis was performed on an EPICS-XL4 (Coulter, France) flow cytometer. In agreement with Crazzolara et al (2001), CXCR-4 was present on the majority of ALL blast cells, as only one case was considered to be negative according to the EGIL criteria. This patient had a favourable outcome. For the 46 cases with positive expression of CXCR-4, the median percentage of positive cells was 87·5% and the MFI was 3·81. No significant difference in CXCR-4 expression was observed between T and B-ALL subtypes. We found a significant correlation between CXCR-4 expression and WBC count at diagnosis using the non-parametric Kendall test (P = 0·01). Patients were divided in two groups according to the presence or absence of splenomegaly and/or hepatomegaly at diagnosis. In agreement with Crazzolara et al (2001), patients with organ infiltration had a significantly higher expression of CXCR-4 than the group without tumoral symptoms (median MFI 4·29 and 2·47, respectively, P < 0·03). We also compared the expression of CXCR-4 on leukaemic cells of patients without relapse, with those who experienced isolated medullary relapse or extramedullary relapse. Surprisingly, no significant difference could be detected between these three groups of patients (Fig 1). We then analysed the expression of CXCR-4 in two groups of children: those with a DFS under 24 months (12 cases) and those with a DFS over 24 months (34 cases). Children with a DFS under 24 months had a significantly higher expression of CXCR-4 (MFI 4·79) than children in first remission at 24 months (MFI 2·85, P < 0·02). Ten patients presented a relapse after 24 months. They had similar levels of CXCR-4 (MFI 2·57) compared with patients in continuous first remission (MFI 2·91). Twelve children died: they had significantly higher levels of CXCR-4 (MFI 4·79) than surviving children (MFI 2·88, P < 0·02). Distribution of CXCR-4 expression on leukaemic cells of children with newly diagnosed ALL. CR: complete remission, IMR: isolated medullary relapse, EMR: extramedullary relapse (associated or not with medullary relapse), DFS: disease free survival. Horizontal lines indicate the median value in each group (▴, deceased children). These results do not confirm that higher levels of CXCR-4 constitute a prognostic factor of extramedullary relapse in children with ALL, but may be associated with a poorer outcome. The reasons of the discrepancy between the two studies are not clear: we used the same antibody to quantify CXCR-4 expression. We cannot exclude an influence of the treatment. In our study, all the children were treated by the French Acute Lymphoblastic Leukaemia Co-operative Group protocol (FRALLE 93), whereas several Berlin–Fankfurt–Munster (BFM) protocols were used in Crazzolara et al (2001). It is notable, however, that the sample size was small in both studies. Despite these restrictions, the data obtained by Crazzolara et al (2001), as well as our data, suggest that CXCR-4 expression analysis in acute leukaemia cells could be of interest. The prognostic relevance of CXCR-4 requires further evaluation in large prospective studies of homogeneously treated patients. This work was supported in part by the Vie et Espoir association.